The kernel has support for setting the baudrate as integer using BOTHER
as id with the termios2 struct. It seems that this is not yet supported
in the libc, so we are using kernel headers directly.

Signed-off-by: Mikhail Vanyulin <mikhail.vanyulin@rtsoft.de>

- '-z' for tx_only mode (yes, we are running out of useful chars...)
- '-r' for repitition now understands '0', to not stop after a counted
  number of cycles.
- Sort the parameters and help, some whitespace cleanups.

Fix for GRNFS-26

Before, the delay_rts_before_send delay_rts_after_send fields were not
explicitly initialized. This led to a erroneous transmission behaviour.
Now the RS485 port settings are read before the half duplex
configuration is done. This ensures that the delays are set to 0 or to
a predefined value (device tree).

BCS 746-000908

If I see it right there could be a segmentation fault if used in echo mode. When you choose echo mode the variable message wouldn't a value assigned and in the cleanup function the not allocated memory would be free'ed.

Option -c enables flowcontrol with rts and cts, for none RS485 mode.
Option -t adds a delay after the reception of the first byte, usable to
test the flowcontrol.

rand was initialized with time(0) which led to the same data if started
twice in the same wall time second.
Now also taking nanoseconds into acount.

BCS 746-000172

The ```gettimeofday()``` function is defined in ```sys/time.h```. Thus
include it.

BCS 746-000125

* Add RS485 echo test for our EMC verification.
* The test is more flexible than the legacy RS485 ping pong test and will
  always recover from transmission errors due to EMI.
* The sent data includes a hard-coded 4-byte header and a 4-byte data length
  field, which is parsed by the receiver. The receiver always scans the
  received data for the header bytes in order to detect a new message.

BCS 746-000125

The rs485pingpon test tool uses partly the tty settings that are setup
prior to execution.
This might lead to problems when the tty is misconfigured.

Use the default tty settings from glibc instead.
This was inspired by "stty sane"

Change-Id: Ie1a5d69edddb967e4dc284a450e3195e53d6af3c

* test.h and others moved to include/old/tetst.h, adapted include path
* gcc7 seems to have stricter tests, fixed some bugs
* refs #30346

Change-Id: I77af8597edaa458bc7376b96a3ec20610db8c3ab

* Moved test tools from recipes-guf to recipes-test
* Removed ac97regtest and pman completely, because they are not relevant for our current platforms anymore
* Corrected some summary entries in bitbake recipes